LLMs are Vulnerable to Malicious Prompts Disguised as Scientific Language

TL;DR

This paper demonstrates that state-of-the-art large language models are vulnerable to malicious prompts disguised as scientific language, leading to increased biases and toxicity, and highlights the need for careful training data considerations.

Contribution

It reveals the susceptibility of LLMs to scientifically disguised malicious prompts and analyzes factors contributing to these vulnerabilities, emphasizing the importance of training data scrutiny.

Findings

Models' biases and toxicity increase with malicious scientific prompts

Models can be manipulated to generate fabricated scientific arguments

Mentioning author names and venues amplifies model biases

Abstract

As large language models (LLMs) have been deployed in various real-world settings, concerns about the harm they may propagate have grown. Various jailbreaking techniques have been developed to expose the vulnerabilities of these models and improve their safety. This work reveals that many state-of-the-art LLMs are vulnerable to malicious requests hidden behind scientific language. Specifically, our experiments with GPT4o, GPT4o-mini, GPT-4, LLama3-405B-Instruct, Llama3-70B-Instruct, Cohere, Gemini models demonstrate that, the models' biases and toxicity substantially increase when prompted with requests that deliberately misinterpret social science and psychological studies as evidence supporting the benefits of stereotypical biases. Alarmingly, these models can also be manipulated to generate fabricated scientific arguments claiming that biases are beneficial, which can be used by ill-intended actors to systematically jailbreak these strong LLMs. Our analysis studies various factors that contribute to the models' vulnerabilities to malicious requests in academic language. Mentioning author names and venues enhances the persuasiveness of models, and the bias scores increase as dialogues progress. Our findings call for a more careful investigation on the use of scientific data for training LLMs.